Data Quality Metrics: Completeness, Validity, Uniqueness, Timeliness & DQ Scorecards

What you'll build

What you will be able to do

Each lab ends with a concrete result that you build yourself and the checker confirms, cell by cell.

• Count the blanks in every column of a customer table, turn them into completeness percentages, measure how many records have all required fields and flag the weak columns.
• Test every email for a basic format, every country against an allowed list and every credit limit for type and range, compute validity rates per rule and per row, and add a Data Validation rule that stops new bad limits.
• Flag rows whose customer ID appears more than once, count distinct IDs, compute the uniqueness rate, find duplicate emails that differ only in capitals and colour the duplicate keys.
• Look up every order's customer in the master table, count the orphan orders and the revenue they carry, find orders whose region contradicts the master and compute a consistency rate.
• Compute how many hours each order took to reach the CRM, the share entered within the 4-hour SLA, the 90th-percentile lag and how stale the whole table is at the reporting time.
• Compare 12 CRM records with values verified at the source, match cities while ignoring case and stray spaces, accept revenues within a 5% tolerance, compute field accuracy and the margin of error of a small sample.
• Combine five dimension scores into one weighted DQ score per dataset, give each a Green/Amber/Red status, name its weakest dimension and add traffic-light icons.
• Compute weekly error rates and the cost of bad records, chart the DQ score with a trendline and estimate in which week the trend reaches the 97% target.